Championship predictions Hex, Havannah

17 replies. Last post: 2020-03-03

Reply to this topic Return to forum

Championship predictions
  • Force majeure at 2020-01-15

    Based on Monte-Carlo simulation, below are top5 rated players and their theoretical chances to win the 23rd Havannah Championship (probability of win in each game is implied from the LG’s  rating formula):70%  eobllor11%  54668%    Force Majeure6%    Mirko Rahn2%    Sol

  • struggler at 2020-01-15

    Cool :P

  • lorentz at 2020-01-15

    Is 5466 a bot?

  • Clark94 at 2020-01-15

    RE: Lorentz.

    Im not sure if 5466 is human or not but I was able to defeat 5466 in the current amazons championship. 5466 does have the highest amazons rating by a big margin however.

  • fritzd ★ at 2020-01-16

    5466 has been around much longer than the NN bots. I’m 99.9% sure he is very much human!

  • Force majeure at 2020-02-21

    Based on Monte-Carlo simulation, below are top5 rated players and their theoretical chances to win the 2 Havannah Championship (probability of win in each game is implied from the LG’s  rating formula):
    Daniel Sepczuk 2443 37.4Arek Kulczycki 2436 37.0lazyplayer         2358 18.6Force majeure 2244 03.9Frode Lillevold 2205 02.0

  • Force majeure at 2020-02-21

    Based on Monte-Carlo simulation, below are top5 rated players and their theoretical chances to win the 2 Havannah Championship (probability of win in each game is implied from the LG’s  rating formula):

    • Daniel Sepczuk 2443 37.4
    • Arek Kulczycki 2436 37.0
    • lazyplayer         2358 18.6
    • Force majeure 2244 03.9
    • Frode Lillevold 2205 02.0

  • KPT ★ at 2020-02-23

    So...basically , the better the ELO greater de chances to win...
    Do you use an excel platform for doing this predictions? I’m trying to understand what is this all about. There is too much stuff in internet...about stocks, investing, horse racing. It is very confusing haha. 

  • Tasmanian Devil at 2020-02-23

    Your probability of winning any game is in theory 1 / (1 + 10^((your rating – opponent’s rating) / 400)). So, using your favorite programming language, you define all the matches within a tournament and generate the outcome at random using this formula. Do this thousands of times and count how many times each player won the tournament.

  • Force majeure at 2020-02-23

    Tasmanian Devil has descreibed exactly what I’ve done :)

  • KPT ★ at 2020-02-23

    I got it.

    what would be the theory for a race competition?where every player is against everyone and not 1vs1

    thanks

  • struggler at 2020-02-23

    Force majeure, I think your predictions concerned the hex championships (not havannah)?

  • Force majeure at 2020-02-23

    Probably :)

  • Arek Kulczycki at 2020-02-28

    I will not be playing any Havannah soon haha!

    Is there really a “400” in that equation???

    Interesting :)

  • Force majeure at 2020-02-29

    Yes there is :)

  • Carroll ★ at 2020-03-03

    It corresponds to a probability of 1/11 for a difference of 400 elo, or the reverse:

    https://www.wolframalpha.com/input/?i=1%2F%281%2B10%5Ed%2F400%29

  • David J Bush ★ at 2020-03-03

    Heh Elo has been “improved” with Glicko, but the graph of rating difference to odds of winning remains about the same. Glicko is more complicated, assigning two numbers to each player: rating and ratings deviation or how accurate your rating is, based on how much you have been playing lately. Then there is Glicko-2, which adds a 3rd value: volatility. It could be argued that the simplicity of Elo is preferable, especially with a relatively small group like us.

    Arek you should definitely play more Havannah and more Twixt.

Return to forum

Reply to this topic




Include game board: [game;id:123456] or [game;id:123456;move:20] or [game;id:123456;move:20;title:some text]